The curious case of the softening spectrum: more on astrophysical neutrinos

IceCube has posted a set of papers on the arXiv, giving new results on starting events: neutrino interactions that occur within IceCube.   These analyses use 102 events observed in 7 1/2 years of data,.  There are many new results, including new measurements of the astrophysical neutrino flux and energy spectrum, evidence for the observation of tau neutrinos and  the  first measurement of the neutrino-nucleon cross-section using starting events.  The papers are available on the arXiv preprint server, and have been submitted for journal publication:

"The IceCube high-energy starting event sample: Description and flux characterization with 7.5 years of data," R. Abbasi et al., available as arXiv:2011.03545.

"Measurement of Astrophysical tau neutrinos in IceCube's high-energy starting events, R. Abbasi et al., available as arXiv:2011.03561.

"Measurement of the high-energy all-flavor neutrino-nucleon cross section with IceCube, R. Abbasi et al., available as arXiv:2011.03560. 

There were a couple of reasons to have three publications.  These are three very different topics, based on rather different analysis techniques.  But, length was also an issue: the first paper comes in at 51 pages, definitely on the long end of the spectrum for physics papers.   This post will focus on the first paper, which also describes the data sample.

The analyses in the first paper are very similar to those in previously published starting event analyses, which I discussed here.   The current analyses benefits from more data, and better detector calibrations and better analysis software, giving better measurements of the energy deposited in the detector, better measurements of the neutrino directions, etc.

That said, the results have changed more than we would have expected.  Most notably, the measured neutrino energy spectrum has gotten softer (i. e. there are fewer very energy astrophysical neutrinos, and more with lower energy).  The figure immediately above shows the energy spectrum (expressed as energy deposited in the detector) and the zenith angle (where cos(theta)=+1 is going vertically downward, and cos(theta)=-1 is going vertically upward), compared to the expectations for atmosphe ric muons, atmospheric neutrinos (labelled as Atmo Conv.) and a fit to the astrophysical spectrum.  The fit found the astrophysical spectrum was consistent with a flux phi=phi_0 (E_neutrino/100 TeV)^-alpha, where alpha=2.87+/-0.20.  Here, phi_0 is a normalization constant.  In comparison, previous contained event analyses found alpha in the 2.3 to 2.6 range, depending on which years of data were studied.    The collaboration spent much time trying to determine what has changed.  Otherwise, this paper would have been out some time ago.  

We looked at every plausible explanation that we could find, and even some that were clearly less plausible.  If we use just the first 4 years of data, the results were similar to those in the previous analysis.  If we swap the old and new software and calibration, very little changes.  There is no evidence for any change in the detector behavior; one expects detectors buried under a mile of ice and held under constant conditions to be very stable, and, as expected, we see no significant changes in atmospheric neutrinos, cosmic-ray muons, or any other measure of detector performance.   The interactions were spread pretty evenly throughout the detector, so it is not a problem in a small part of the detector.   The astrophysical neutrinos come from a large number (very likely >50) source from many directions in the sky, so it is not plausible that this is due to a change in their source.   So, in the end, I am just chalking this up to statistics - once in a while, we expect large (roughly 2 sigma) statistical fluctuations, and this seems to be one of those occasions.

 The neutrino arrival directions have also changed somewhat.  This is better understood, and comes from a combination of improved analysis techniques and a better understanding of how light scatters and is absorbed in the Antarctic ice.  For each neutrino candidate, we estimate the probability of it coming from any given direction in the sky.  The result is a blob (which may be regular or irregular, depending on the reconstruction) centered around the most likely arrival direction.  The graphic at the top shows our revised sky map, which shows the estimated flux coming from different directions, where we add up the probability that each neutrino came from a given direction.   The gray dot shows the center of our galaxy, and the gray curve shows the galactic plane.

The color code gives the "Test statistic," a measure of how likely the measured flux from that direction can be explained by background.   There is a hot spot (every map must have a hottest spot), but it is not statistically significant; this map shows no evidence for any specific neutrino sources.  It should be noted that, because we have only a handful of contained events, this search is less sensitive than studies using through-going muons.

 

Science in the age of Covid

 Hi,

Apologies because I have not updated this blog in quite a while.   I'm healthy, but Covid has brought many changes to my work life, and almost everything takes longer.    

Covid has had an enormous effect on scientists, and a somewhat lesser but still very sizable effect on science.  I am one of the fairly large fraction of physicists who mostly work on a computer.  So, I am able to work at home, and I do so.  There is a significant hit to productivity because I can no longer walk next door and talk to my colleagues - everything requires an appointment and a zoom call. 

Laboratory science has taken a much larger hit.  Lawrence Berkeley Lab, like most other research institutions, was almost entirely shut down for about six months, so all laboratory work stopped.  Now, we are slowly and carefully ramping up lab work, with a whole host of anti-Covid precautions, about physical separation, etc.   Of course, Covid-related work has become very high priority.  At LBNL, this includes using the Advanced Light Source (it produces intense beams of X-rays) to study the structure of important proteins, using the NERSC supercomputers to study protein-Covid interactions, and, of course, much biological research. 

Polar science has been affected even more.  There is a strong determination to keep Covid out of Antarctica.  To do so, the U. S. polar program has slashed the number of people who are going there this coming Astral summer.   The only activities that are supported are those that keep the U. S. stations running, and prevent damage to scientific infrastructure.   For IceCube, we will be able to swap winter-overs as the Pole, but not much more.   The usual transportation, using Air Force and NY Air National Guard LC-130 transports will not occur; instead there will be a small number of flights on Baslers (shown above, they are much upgraded DC-3s with turboprop engines), with very limited passenger space, and even more limited cargo transport.  

Everyone going to Antarctica will spend time quarantining in both the U. S. and New Zealand.  Even this has been tough - New Zealand is essentially closed to visitors, so special arrangements were required to allow polar program personnel in.   Fortunately, IceCube is running well, so the main effect is to put off some planned software upgrades, plus the surface deployment of new prototype air shower detectors.

For younger scientists without career positions,, the effects of Covid are especially drastic.  Colleges and universities are in dire straits financially, and have consequently cut faculty hiring.  One estimate I saw was that the number of advertised faculty positions is down by 70%.    This is a huge cut, especially for people who were positioning themselves to apply for jobs this year.  The situation in industry is better, but it is still not as good as last year.    Overall, physicists are probably no worse off than most other professions, but young people have a limited time window to apply for faculty positions, and budget limitations will create a squeeze that will likely last for several years.

Here comes the tau?

Simulated tau neutrino event in IceCube.  Each sphere is an optical module that observed light, with the size scaling with the number of photons.  The color indicates the timing of the light, from red (earliest) to blue (latest).

 One of the more interesting/embarrassing holes in IceCube's physics portfolio was the tau neutrino.  Neutrinos come in three flavors: electron, muon and tau, each tied to the charged lepton of the same name.  Over long distances, these neutrinos can oscillate, changing flavors.  So, no matter what flavor ratio a neutrino beam is produced with, over long distances, we expect it to oscillate and reach Earth as a roughly 1:1:1 ratio of electron:muon:tau neutrinos.

Since very few tau neutrinos are produced directly on Earth, the observation of tau neutrinos was considered to be a clear sign of astrophysical neutrinos, and many many papers discussed the signatures and expectations.  A beautiful 1995 paper by John Learned and Sandip Pakvasa   (also available on the arXiv - soon after it was founded) pointed out that sufficiently energetic tau neutrinos could produce a distinctive 'double bang' signature - a large cascade when the neutrino interacted, and a second when the resulting tau lepton decays.  Even though the tau lepton lifetime is very small (3*10^-13 s), when it has energies of a PeV (10¹⁵ eV) or higher, a Lorentz boost extends its lifetime in the Earth frame of reference so that the two bangs can be separated by an average of (Energy/1 PeV) * 50 meters, making for a distinctive signature seen in the simulation shown above, with two distinctive light clusters.  Unfortunately, IceCube has not seen this signature, and we have also not seen enough PeV-energy neutrinos so that we can expect to see it.

A candidate tau neutrino event seen in IceCube. Each sphere is an optical module that observed light, with the size scaling with the number of photons.  The color indicates the timing of the light, from red (earliest) to blue (latest).  The seven plots show the waveforms (light vs. time) for certain optical modules; several show apparent double-pulse signatures.

However, IceCube is developing techniques that will allow us to see tau neutrinos with lower energies, where the two bangs are closer together.  Even if they are so close together (10-30 m) that we cannot separate the overall light clouds,  there may be some optical modules that see pulses from the two cascades at separate times, producing a double-pulse topology in an individual optical module.   The figure above shows one candidate event, along with waveforms from some of the modules, showing the double-pulse signature.   A word of caution is in order - there are some possible background processes that could mimic these signatures - but this is considered by IceCube to be evidence for tau neutrinos.  "Evidence" typically means that the statistical significance is 3 sigma or more, not the 5 sigma required to claim a detection.  Since we expect to see tau neutrinos, this is reasonably convincing, an it seems safe to say that the holes has largely been filled in.  We look forward to more precise measurements, of course, to check in more detail for consistency with the standard acceleration scenarios.

The tau neutrino work has been presented in several recent conference presentations, including ones with writeups by Daan van Eijk and Logan Wille and Juliana Stachurska.

More neutrino interaction physics with IceCube

Once again, IceCube has shown that we can study high-energy neutrinos in their own right, rather than just as astrophysical probes.   This analysis used a sample of starting tracks in 5 years of data, from neutrinos that interacted within the detector, producing a hadronic cascade from the nuclear target recoil, and a muon from the lepton, in a reaction written as neutrino + nucleon (proton or neutron) -> muon + X, where X is the shower of particles produced by the recoiling nucleon.   In these interactions, there are two quantities to measure, the energy of the muon, and the energy of the shower.  The inelasticity is the energy of the cascade divided by the total energy (the sum of the shower and muon energy).  The distribution of inelasticity is well predicted by the Standard Model of Particle Physics, but has not been measured at energies above 500 GeV (5*10^11 electron volts).  With IceCube, we have extended the measurement to energies above 100 TeV (10^14 electron volts) - a factor of 200 upward in energy.  This plot shows the measured average inelasticity, from our new Icecube preprint, available here, or directly as pdf. 

The points show the inelasticity, while the blue and green curves show the standard model predictions for neutrinos and antineutrinos respectively.  The red curve shows the expectation for the mixture expected in IceCube.  For aficionados, this calculation is done at next-to-leading order accuracy, with BFKL evolution to low-x partons.

This measurement is sensitive both to potential beyond-standard-model physics, which would likely have a rather different inelasticity distribution than for the expected interactions.  Even the standard model cross-section is sensitive to the number of low-momentum quarks and antiquarks in the target nucleus.

Inelasticity is interesting in it's own right.  But, the inelasticity can also be used to probe a number of additional physics topics.  Neutrinos and antineutrinos have different inelasticity distributions, so by assuming the standard model values, we can measure the neutrino:antineutrino ratio.  As can be inferred from the plot above, it is exactly as expected.  Unfortunately, at the energies where IceCube is sensitive, we are mostly studying atmospheric neutrinos, not astrophysical.

We can also use inelasticity to probe astrophysical neutrinos.  Although the neutrinos selected here are mostly muon neutrinos, some tau neutrinos make it into the fit, and it is possible to use similar criteria to select a matching set of cascades.  The plot below shows the flavor triangle found from this study.

 Each point on the flavor triangle corresponds to a unique mixture of electron, muon and tau neutrinos.  The upper point is all muon neutrino, with the lower left and lower-right points corresponding to all tau neutrinos and all electron neutrinos respectively.  The colors show the relatively likelihood, with the best-fit point (cross) corresponding to 83% tau neutrino, 17% electron neutrino and no muon neutrino.  Unfortunately, the errors are large, so none of the different standard acceleration scenarios can be ruled out.

This work was done by my student, Gary Binder.  Besides the IceCube paper, he wrote a very nice dissertation, available here.  For this work, he won the GNN (Global Neutrino Network) dissertation prize for 2018. 

Found: one cosmic accelerator - TXS0506+56

Yesterday, in two papers published in Science, IceCube and collaborating experiments announced the observation of high-energy astrophysical neutrinos coming from a source, the blazar TXS0506+56. (the numbers denote its position in the sky).  This was announced at a press conference at National Science Foundation headquarters, and accompanied by press releases from multiple institutions, including from one from Berkeley Lab, and made the cover of Science (above).

Blazars are a type of active galactic nuclei (AGNs), which are themselves galaxies with a supermassive black hole at the center.  If the black hole is surrounded by a dust cloud (or other matter), it will gradually accrete that matter.  In the process, it will eject a fraction of it in a relativistic jet perpendicular to the galaxies axis of rotation.  The jet is turbulent, and thought to be a likely site to accelerate particles to extremely high energies.  In blazars, this jet is pointed nearly directly at Earth, giving us the best chance to see these ultra-energetic particles. 

The story begins on Sept. 22, 2017, when IceCube observed a neutrino with an energy around 300 TeV (about 50 times the energy of the protons accelerated at CERNs Large Hadron Collider).   The event display shows the neutrino; each colored dot shows the photons registered by one IceCube optical module, with color indicating relative time (from red to blue), and the size indicating the number of photons.

IceCube has seen many neutrinos that were more energetic than this, but this was still enough energy that the neutrino was likely of astrophysical origin.  So, computers clacked and whirred, and 43 seconds later we sent out an automatic alert, telling many partner observatories that we had seen an energetic neutrino coming from a specific direction.  Several of these observatories pointed telescopes in the direction of the neutrino, and (to make a long story short) that the location coincided with a minutes to years; this neutrino came during a time when TXS0506+56 was emitting at particularly high levels, from radio waves through (at least) 1 TeV photons.  The  first paper, by IceCube and the other experimental collaborations, discusses this coincidence in space and time.   TXS0506+56 is a relatively energetic, quite nearby (with a redshift of 0.3365), so it is a likely candidate for a first observation. 

Shortly after this observation, IceCube went back and looked at archival data, searching for excess emission from the source.    We found an excess of neutrino events coming from that direction during the period from September 2014 to March 2015.   This is reported in the second paper.

The exact statistical significance of these observations depends on some of the details of the analysis - the first paper gives a range of significances, depending on the preferred assumptions. But, taken together, this is strong evidence that we have seen neutrinos coming from a specific source: we have found at least one cosmic accelerator, far more powerful than CERN's LHC.   Besides the observed neutrinos, there is strong suspicion that AGNs also accelerate the ultra-energetic protons and/or heavier nuclei cosmic-rays that led us to look for neutrinos in the first place.  Unfortunately, since protons and heavier nuclei are bent by interstellar magnetic fields, they do not point back to their sources.

One still-open question is whether blazars are responsible for all of the neutrinos that IceCube sees.  In 2016, IceCube published a paper (freely available arXiv version here) which set limits on the fraction of the astrophysical neutrino that could come from blazars, setting a limit between 27% and 50%, depending on the spectral index.  This paper studied 862 blazars, and had to make some assumptions about the relationship between the observed gamma-ray flux and the expected neutrino flux.   As you can imagine, extensive work is ongoing to revisit this question.

Stopping a neutrino beam; measuring their interaction cross-section

Neutrinos are popularly known as the particles that go through anything and everything.  Neutrinos from beta decay can escape from the best shielded nuclear reactor, and neutrinos from nuclear fusion escape from the center of the sun.  Neutrinos interact only via the weak interaction, which is indeed weak.  But, that doesn't mean that they can go through anything - the IceCube Neutrino Observatory recently demonstrated experimentally that it is possible to stop a beam of neutrinos, in a paper published in Nature (also freely available on the arXiv).

To do this, IceCube used two tricks. 

First, it use extremely energetic neutrinos, with energies above 1 TeV (1 tera-electron volt, or 10¹² electron Volts), extending up to 1 PeV (1 peta-electron volt, or 10¹⁵ eV), millions of times more energetic than neutrinos from nuclear fusion or radioactive ion decay.  The cross-section (probability) for neutrinos to interact rises with energy (linearly at first, then moderated to scale roughly as Energy^0.3.  So, at an energy of 30 TeV (the rough mid-point of the measurement) the cross-section is several million times higher than it is for neutrinos from radioactive decay.  Of course, there aren't that many neutrinos this energetic, but, at 1 cubic kilometer in volume,  IceCube is big enough to collect a good sample.  The analysis used 10, 784 energetic muons from neutrinos that passed through at least some of the Earth.

Second, it used a very thick absorber - the Earth.  With this, the measurement was quite simple.   It Compared to a baseline of near-horizontal neutrinos that traversed only a relatively small amount of matter, energetic near-vertical neutrinos were absorbed going through the Earth.  The figure above shows the predicted transmission probability (= 1 - absorption probability), as a function of neutrino energy and zenith angle; the latter shows how much Earth matter was traversed.   

There are of course many complications - experimental uncertainties on the neutrino energy, neutral current interactions, where a neutrino may emerge from the Earth with a lower energy than it entered, modelling the material within the Earth, etc., but the result clearly showed that neutrinos are absorbed at about the expected rate.  More precisely, the best-fit cross-section was. 1.3 +/- 0.5 times the predictions of the Standard model where I have combined the statistical and systematic uncertainty.  It was not trivial to find a good definition for the neutrino energy range for which this measurement applies, because different methods give somewhat different energy ranges, but we settled on a method that returned a range from 6.3 TeV to 980 TeV.  For comparison, the highest energy measurements at an accelerator laboratory only reached 0.37 TeV - our measurement reaches order of magnitude higher energies than than.  The figure below puts this in perspective, comparing our measurement with the previous accelerator work.  The cross-sections (y axis) are divided by the neutrino energy so that everything fits on the graph better; otherwise, it would span many orders of magnitude.

I have to mention that this was the dissertation work of my (now graduated) graduate student, Sandra Miarecki.  Sandy had a very interesting preparation for graduate school - she was a career US Air Force Pilot, serving many roles, including as a test pilot, before retiring from the Air Force and coming to graduate school in Berkeley.   After graduate school, she became an Assistant professor at the US Air Force Academy.   The LBNL news center has a very nice article about her.

The Nature article also recieved a fair amount of press coverage.  I will just mention one article,  in Symmetry magazine, which goes into more detail about the analysis than other press writeups.

Gravity waves, Gamma-rays and gold jewelery

It has been a bumper month for astrophysicists.

On October 16th, the combined LIGO/VIRGO collaborations announced the observation of gravitational waves from an even that occurred on August 17th.  Unlike the previous observations, these waves came from relatively 'light' objects, reflecting the collisions of two presumed neutron stars, with masses around 1.1 to 1.6 times the mass  of the sun, forming a black hole with a mass around 2.74 times the mass of the sun.   Previous gravitational wave events had come from the collisions of much heavier objects.

But, that's not all.  Two seconds later, the FERMI observatory, a satellite containing a large gamma-ray detector, and the INTEGRAL satellite both observed pulses of gamma-rays coming from the same direction.   This is the classical signature of a 'gamma-ray burst' (GRB).  GRBs were first observed in the 1960's by the VELA satellites, built to monitor gamma-rays from possible atmospheric or space-based nuclear weapons tests.   VELA did not observe these, but it did find mysterious bursts of gamma-rays coming from space.     These bursts have been the subject of scientific speculation for decades, and the conventional wisdom was that some GRBs came from the merger of neutron stars or black-hole on neutron star mergers.  That theory has now been amply confirmed by the LIGO/VIRGO/FERMI/INTEGRAL observation.   The graphic above, from the LIGO collaboration, shows the process.

Of course, this collision site was studied by many many other astronomical instruments.  IceCube looked, but we didn't see anything.   However, the optical studies were very fruitful.  Multiple telescopes observed an optical signal that lasted for a few days, plus an infrared signal that lasted for nearly two weeks.  These signals were consistent with some predictions made by my LBNL colleague Dan Kasen and his collaborators.  Kasen made a detailed model of the graviational, nuclear and atomic processes that would occur in a collision of two neutron stars, and, from that, predicted the optical and infrared light emission.  His model predicts considerable production of heavy elements (heavier than iron) via rapid neutron capture (the 'r-process').   The shorter-lived broadband optical emission comes from an initial ejection of lighter nuclei. The long-lived infrared component comes from a secondary emission which is powered by the radioactive decay of heavy elements which heat the plasma that surrounds the newly formed black hole.  Heavy elements (Z between 58 and 90) scatter the light strongly, so it takes longer to escape from the plasma.

This agreement is of great interest to nuclear physicists, since it may provide a new answer to the question: where do the heavy elements in the universe come from?  Previously, it was thought that they were mostly produced in supernovae, explosions that occur when heavy stars reach the end of their livetime and collapse.  However, Dan's simulations  shows that GRBs produce heavy elements, and could account for much or all of the gold used in our jewelry, along with all of the other heavy elements.